I find it a little hard to believe that they're inaccurate. However, you could try writing your own (probably slower) sin, and arcsin routines from which you can get most of the others.
You could use a Taylor series expansion, for example. Keep in mind that you only need to deal with the first pi/2 radians (the first 90 degrees) as everything else is a reflection or a shift.
Might I ask how you determined that sin and cos weren't being accurate? They should be accurate to a pretty large number of decimal places.
It looks like you're new here. If you want to get involved, click one of these buttons!